{"id":"5e9e1c78-c95c-4ddb-931d-728b49ab2f94","arxiv_id":"2508.18987","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":9,"one_line_summary":"Ornstein-Uhlenbeck fluctuations in pairwise adhesions fluidise a multi-phase field epithelial monolayer and produce a non-monotonic dependence of cell diffusion on the fluctuation persistence time.","lead":"Junction-level fluctuations in cell adhesion can fluidise a simulated epithelial sheet, turning it from a solid-like hexagonal lattice into a moving, disordered tissue. The paper shows, in a multi-phase field model, that this works best for an intermediate persistence time of the fluctuations, matching earlier results from vertex models.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Non-monotonic D_diff(τω) may be a finite-time/statistical artifact: single runs and T comparable to largest τω; replicate and longer runs needed.","rationale":"The reader's weakest assumption—that the non-monotonic dependence is a steady-state property and not a finite-time artifact—is exactly the point I would stress. The paper is honestly written and explicitly flags the finite-time limitation, and the SI provides supporting evidence (diffusive MSD exponents, non-monotonic trend for τω<10^4). However, the absence of error bars or multiple independent runs means the ordering of D_diff values across τω could be statistical noise, and the large-τω end is affected by short equilibration relative to τω. This is not a claim of internal inconsistency; it is a request for stronger evidence on a quantitative, load-bearing element of the paper. The mechanism itself—junctional fluctuations inducing T1 transitions and fluidisation—is physically plausible and supported by the qualitative solid-to-fluid crossover. I see no reason to change the reader's CONDITIONAL verdict; the paper should be published with the caveat that the non-monotonic dependence needs confirmation with replicated runs and longer simulations. The proposed concrete test would settle whether the concern actually lands.","tokens_in":9362,"tokens_out":5376,"duration_ms":51908,"concrete_test":"Pick 4 points spanning the non-monotonic peak at σω=1.5 (τω=10^2, 10^3, 10^4, 10^5). Run N≥10 independent seeds per point; compute mean and standard error of D_diff from both MSD(T)/(4T) and a linear fit to MSD(t) over t∈[T/2,T]. Also run one simulation at τω=10^5 with T=10^6 and with the OU process pre-equilibrated for 5τω before measurements. If the non-monotonic peak survives (D at τω=10^5 significantly below peak, with non-overlapping error bars) and the longer run does not raise D at τω=10^5 above the τω=10^3 value, the finite-time/statistical concern is ruled out; otherwise the central claim needs qualification.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section III defines D_diff ≡ MSD(T)/(4T) with T=1.8×10^5 and reports heatmaps (Fig. 3a) from what appear to be single simulations per (τω,σω) point. The authors explicitly state 'the final simulation time is comparable with the highest values of τω we consider' (end of Section III). This matters because for τω=10^5, the measurement window is only ~1.8 τω, and the pre-measurement equilibration of 3×10^4 is only 0.3 τω. If ω_ij is initialized at zero (not stated), the OU process has not reached its stationary variance during the measurement window, so D_diff at large τω could be suppressed by transient low adhesion, producing the observed downturn. The SI mitigates this by showing non-monotonicity for τω<10^4 and near-linear MSD at τω=10^4 and 10^5, but each point remains a single trajectory; no error bars are given. The load-bearing premise is therefore not just 'simulation time is long enough' but 'MSD(T)/(4T) from one run is a reliable estimator of the asymptotic diffusion coefficient across this τω range.' This is the weakest point of the otherwise plausible mechanism.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper adapts the Ornstein-Uhlenbeck junctional-fluctuation mechanism, previously used in vertex models, to a multi-phase field model of an epithelial monolayer. Pairwise cell-cell adhesions ω_ij are made to fluctuate stochastically (Eq. 4d and Eq. 5), and the authors show, via a hand-tuned T1 transition (Fig. 1) and via scans of the variance σ_ω and persistence time τ_ω, that sufficiently strong fluctuations fluidize the monolayer: the MSD becomes diffusive, neighbor-change events occur at a steady rate, and hexatic order is lost (Figs. 2 and 3). The effective diffusion coefficient D_diff = MSD(T)/(4T) is reported to depend non-monotonically on τ_ω, consistent with a vertex-model result [Yamamoto et al., Soft Matter 18, 2168 (2022)]. The authors explicitly acknowledge that the total simulation time is comparable to the largest τ_ω values, and they provide supporting MSD exponent fits in the Supplemental Material.","tokens_in":9741,"tokens_out":6533,"duration_ms":64483,"significance":"If the claims hold, the paper adds a new, physically motivated fluidization mechanism to the multi-phase field modeling toolkit, analogous to the well-established fluctuating-tension approach in vertex models. The non-monotonic dependence of D_diff on τ_ω, if robust, is an interesting generic feature worth reporting. The work is transparent: the model is fully specified, the parameter values are given, and the SI contains additional MSD data and T1 movies. The main contribution is computational, not analytical, and the authors do not overclaim beyond their simulation evidence.","major_comments":[{"comment":"The central quantitative claim, the non-monotonic D_diff(τ_ω), rests on single simulations per parameter point. D_diff is defined as MSD(T)/(4T) at T=1.8×10^5, and no error bars, replicate runs, or statistical significance tests are reported. In a driven 2D tissue, MSD(T)/(4T) from one trajectory can fluctuate substantially, especially near the solid-fluid transition. Please provide at least several independent runs (different noise seeds) per parameter point, report mean ± SEM, and confirm that the non-monotonic shape survives averaging. Without this, the non-monotonicity may be a statistical fluctuation.","section":"Section III, Eq. (6) and Fig. 3(a)"},{"comment":"The initialization of ω_ij is not stated. If, as suggested by the text ('when a pair of cells newly come into contact their adhesion is zero'), ω_ij starts at zero, then for τ_ω=10^5 the OU process has not reached stationarity: the 3×10^4 equilibration is only 0.3 τ_ω, and the variance is still increasing during the measurement window (T=1.8 τ_ω). This can suppress rearrangements at large τ_ω and create an artificial downturn in D_diff. The SI MSD exponents show diffusive motion at τ_ω=10^5, but they do not establish steady state. Please initialize ω_ij from the stationary Gaussian distribution, and/or extend at least the large-τ_ω simulations well beyond several τ_ω to demonstrate that the downturn is not a finite-time artifact.","section":"Section III, last paragraph; Eq. (5)"},{"comment":"The summation convention in Eq. (4d) is ambiguous. The sum over i and j∈N_i(t), with ω_ij=ω_ji, counts each unordered pair twice unless a 1/2 factor or an ordered restriction is explicitly stated. The Supplemental Material's functional derivative, δF_adh/δφ^(i) = -4∑ω_ij λ φ^(i) ∇²[(φ^(j))²], is consistent with this double counting. If instead the intended adhesion energy is a sum over unordered pairs, Eq. (4d) and Eq. (S2) both contain a factor-of-2 error, which would change the effective relaxation rate J0 and shift the phase boundary. Please state the convention explicitly and, if double counting is intended, note it explicitly because it changes the energy scale by a factor of 2 relative to the single-pair convention used in some previous phase-field work.","section":"Eq. (4d) and Supplemental Eq. (S2)"}],"minor_comments":[{"comment":"Please clarify whether N_i(t) includes i itself; if not, the condition ω_ij=ω_ji and the restriction j∈N_i(t) are enough to define a well-defined pairwise term.","section":"Section II, Eq. (4d)"},{"comment":"Minor wording: 'Exponents of a MSD(t)=αt^β fit' should be 'fits' (plural); also please state the fitting range used for the exponent.","section":"Supplemental Material, Fig. S2 caption"},{"comment":"The color bars in Fig. 3 are helpful, but the reader cannot tell from the heatmap how many grid points are used in each direction or whether any intermediate values were interpolated. Please indicate the discrete parameter values explicitly (e.g., markers on the axes).","section":"Section III, Fig. 3"},{"comment":"The claim that the modified adhesion term (4d) is 'less likely' to produce large forces in low-φ regions is plausible but not quantified. A brief test of the maximum stable ω_ij for the old and new terms would make the numerical-stability statement more concrete.","section":"Supplemental Material, stability discussion"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope for a soft-matter/biological physics journal and the central mechanism is plausible. My main concern is statistical: the non-monotonic τ_ω dependence, which is the headline result, is supported by single runs and by a measurement window comparable to the largest persistence times. This is fixable with added simulations and a clearer statement of the OU initialization, so I recommend major revision rather than rejection. The summation-convention ambiguity in Eq. (4d) should also be resolved before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper does what it says – OU fluctuations in pairwise adhesions fluidise a multi-phase field monolayer, and the effective diffusion coefficient depends non-monotonically on persistence time, matching vertex-model results. The modelling is clean and the SI actually rebuts your biggest worry.\n\nWhat's new: first OU-flavored junctional noise in a phase-field model, plus a modified adhesion functional whose derivative scales as φ^(i), which the authors argue improves stability at large adhesion. Both are minor but real technical increments. The demonstration that a single T1 can be triggered by modulating ω_ij is nice and clearly presented.\n\nCredit where due: the fluid/solid contrast at σ_ω=0.1 vs 1.1 is stark and convincing. The neighbour-change statistics and hexatic order corroborate the MSD. And the authors explicitly head off the finite-time critique: the non-monotonic trend is present for τ_ω<10^4, with simulation time an order of magnitude larger, and the log-log MSD fits at τ_ω=10^4 and 10^5 are diffusive. So your stress-test concern about the downturn at τ_ω=10^5 being an artifact doesn't threaten the central claim; it only says the rightmost heatmap points shouldn't be trusted quantitatively.\n\nSoft spots: every parameter point is a single run, no error bars anywhere. D_diff is defined as MSD(T)/(4T), a single-time estimator rather than a fit to the diffusive regime; fine when MSD is linear, but risky near the transition. The initial value of ω_ij is never stated – if the OU process starts from zero, large-τ_ω runs spend a meaningful fraction of the measurement window approaching the stationary variance. Again, this only affects the largest τ_ω values, not the peak. No code or data are made available, which is a pity for a methods-oriented paper.\n\nVerdict: this is a useful, honest simulation paper. The central observation is robust in the parameter range where the simulation time is safely converged. It deserves peer review, and I'd send it out as-is with a request for error bars and at least one replicate at the key parameter points. The authors should also state their ω initialisation. For the phase-field community, it's a practical fluidisation knob worth knowing; for the vertex-modelling people, it's a validation that their mechanism carries over. I would bring it to a reading group and would cite it if I were developing phase-field models of tissue. Recommendation: accept peer review; expect a conditional at most, mainly on reproducibility.","headline":"Useful and honest simulation study; the claimed non-monotonicity holds for persistence times safely below the simulation window, so the main finite-time worry doesn't land.","tokens_in":10125,"tokens_out":3270,"would_cite":true,"duration_ms":29603,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Stochastic adhesion at cell junctions fluidises a phase-field epithelial model.","keywords":["epithelial monolayers","multi-phase field model","junctional adhesion fluctuations","Ornstein-Uhlenbeck process","T1 transitions","tissue fluidisation","cell diffusion","hexatic order parameter"],"falsifier":"Repeat the parameter scan with independent realisations and run times several times longer than tau_omega, for example T > 10 tau_omega at tau_omega = 1e5. If the effective diffusion coefficient stops being non-monotonic in tau_omega, or the mean-square displacement becomes subdiffusive on the longer window, the claimed fluidisation would be a finite-time artefact. Alternatively, directly measuring the rate of T1 barrier crossings as a function of tau_omega should show a peak at finite tau_omega if the separation-of-timescales explanation is right.","tokens_in":9299,"feed_emoji":"🔬","tokens_out":5751,"duration_ms":50739,"temperature":0.7,"pith_summary":"This paper asks whether random, time-varying strength of cell-cell adhesion alone can make a model epithelium behave like a fluid. The answer it argues is yes: when pairwise adhesion strengths between neighbouring cells fluctuate as Ornstein-Uhlenbeck processes, the tissue undergoes repeated T1 neighbour exchanges and cells move diffusively, whereas constant adhesion leaves the layer jammed in a hexagonal solid. The central quantitative claim is that the effective diffusion coefficient rises with the fluctuation amplitude but depends non-monotonically on the fluctuation persistence time: there is an optimal persistence time at which cell diffusion is fastest. This matters because it tests, in a phase-field representation of cells, a fluidisation mechanism previously established only in vertex models, and it offers a route to tune tissue rheology through junctional noise alone.","feed_headline":"Junction noise alone makes model epithelial cells flow","feed_subtitle":"Randomly varying cell-cell adhesion triggers rearrangements, and cell diffusion peaks at an intermediate persistence time.","key_machinery":"The central object is the pairwise adhesion coefficient omega_ij(t) appearing in the adhesion free energy F_adh = sum omega_ij lambda integral grad[(phi_i)^2] dot grad[(phi_j)^2]. Each omega_ij evolves by d omega_ij/dt = -omega_ij/tau_omega + xi_ij(t), with Gaussian white noise of variance 2 sigma_omega^2/tau_omega; this is the junctional-fluctuation mechanism, carried over from vertex models. The work it does: transiently strong adhesion pins neighbours together and transiently negative adhesion pushes them apart, driving the fourfold-vertex and T1 intercalation events that let cells exchange neighbours and diffuse. Its functional derivative is proportional to phi_i, a modification of earli","core_discovery":"The paper's discovery is that a multi-phase field model of an epithelial monolayer is fluidised by Ornstein-Uhlenbeck fluctuations of the pairwise adhesion energies omega_ij between neighbouring cells. With small fluctuation amplitude the tissue stays solid and hexagonal with no neighbour exchanges; with large amplitude cells undergo T1 intercalations, neighbour turnover, linear mean-square displacement, and loss of hexatic order. Scanning the persistence time tau_omega and variance sigma_omega^2, the authors find that the effective diffusion coefficient and neighbour-change rate are non-monotonic functions of tau_omega, with an intermediate persistence time maximising mobility, in agreement","pith_inferences":["If the non-monotonicity is generic, junctional-noise-driven fluidisation should also appear in phase-field models with different adhesion functional forms, such as interface-length-dependent adhesion, as long as the effective barrier to T1 events is modulated on a comparable timescale.","The paper's single-run parameter scan leaves open whether the optimal tau_omega depends on system size, area constraint, or friction; a testable extension is to measure the peak position as these are varied.","In real tissues, junctional myosin turnover has a timescale; the model suggests a tissue could be fluidised or solidified by shifting only that timescale, for example through biochemical perturbation, without changing average adhesion strength.","Because the diffusion coefficient is extracted from MSD(T)/(4T) with T comparable to the largest tau_omega, longer-time simulations with multiple realisations would directly test whether the reported peak is steady-state behaviour rather than a slow crossover."],"forward_implications":["In the multi-phase field model, junctional adhesion noise alone can produce a solid-to-fluid transition; no cell-scale polarity or nematic activity is required.","The effective diffusion coefficient and neighbour-change frequency both peak at an intermediate adhesion persistence time, so there is an optimal correlation time of junctional turnover for tissue fluidisation.","The T1 events driven by fluctuating adhesions account for the diffusive mean-square displacement, ruling out solid flocking as the source of cell mobility.","Fluidised and solid phases can be distinguished by hexatic order |psi_6|, which drops from about 1 to about 0 as sigma_omega and tau_omega increase."],"supporting_citations":[{"why":"Reports the vertex-model result this paper reproduces: cell diffusion is non-monotonic in junctional-tension persistence time.","marker":"[21]"},{"why":"Introduces Ornstein-Uhlenbeck tension fluctuations in vertex models and shows they induce tissue fluidisation via T1 transitions.","marker":"[18]"},{"why":"Uses fluctuating junctional tensions in vertex-model cell sorting; supplies the correspondence between the adhesion term and vertex line tension.","marker":"[19]"},{"why":"Analyses how fluctuating tensions overcome T1 barriers, the mechanism the phase-field T1 demonstration relies on.","marker":"[20]"},{"why":"Shows the fluctuating-tension module extends to nematic junctional forces, motivating the transfer to phase-field models.","marker":"[25]"},{"why":"Provides the multi-phase field equations of motion and free energy that the model modifies.","marker":"[29]"},{"why":"Earlier phase-field model with adhesion and polar activity; its active fluidisation is the baseline that must be distinguished from junctional fluidisation.","marker":"[32]"},{"why":"Supplies the numerical method used to solve the phase-field equations.","marker":"[33]"},{"why":"Supplemental Material; provides the finite-time checks (MSD exponents, early-tau subrange) that defend the non-monotonic claim against run-length artifacts.","marker":"[35]"}],"fun_headline_variants":["Junction noise alone makes model epithelia flow","Intermediate junction noise maximizes cell diffusion","Junctional fluctuations fluidize epithelial monolayers","Cell mobility peaks at intermediate junction persistence time","Noisy cell adhesions trigger tissue rearrangements"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The paper's central non-monotonic result is only as strong as its assertion that the measured diffusion coefficient is a converged steady-state property: the simulations run for a time comparable to the largest persistence times considered, with one realisation per parameter point and no error bars, so the outcome would collapse if longer or repeated runs changed the trend.","fun_headline_variants_meta":{"raw":{"variants":["Junction noise alone makes model epithelia flow","Intermediate junction noise maximizes cell diffusion","Junctional fluctuations fluidize epithelial monolayers","Cell mobility peaks at intermediate junction persistence time","Noisy cell adhesions trigger tissue rearrangements"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000191,"raw_usage":{"total_tokens":1114,"prompt_tokens":614,"completion_tokens":500,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":358,"completion_tokens_details":{"reasoning_tokens":434}},"tokens_in":358,"tokens_out":500,"duration_ms":5931,"temperature":1.0,"reasoning_tokens":434,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T16:02:27.128098+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Repeat the parameter scan with independent realisations and run times several times longer than tau_omega, for example T > 10 tau_omega at tau_omega = 1e5. If the effective diffusion coefficient stops being non-monotonic in tau_omega, or the mean-square displacement becomes subdiffusive on the longer window, the claimed fluidisation would be a finite-time artefact. Alternatively, directly measuring the rate of T1 barrier crossings as a function of tau_omega should show a peak at finite tau_omega if the separation-of-timescales explanation is right.","supporting_citations":[{"cited_title":"Yamamoto, D","cited_arxiv_id":null,"evidence_quote":"Reports the vertex-model result this paper reproduces: cell diffusion is non-monotonic in junctional-tension persistence time."},{"cited_title":"Curran, C","cited_arxiv_id":null,"evidence_quote":"Introduces Ornstein-Uhlenbeck tension fluctuations in vertex models and shows they induce tissue fluidisation via T1 transitions."},{"cited_title":"Krajnc, Soft Matter 16, 3209 (2020)","cited_arxiv_id":null,"evidence_quote":"Uses fluctuating junctional tensions in vertex-model cell sorting; supplies the correspondence between the adhesion term and vertex line tension."},{"cited_title":"Duclut, J","cited_arxiv_id":null,"evidence_quote":"Analyses how fluctuating tensions overcome T1 barriers, the mechanism the phase-field T1 demonstration relies on."},{"cited_title":"Duclut, J","cited_arxiv_id":null,"evidence_quote":"Shows the fluctuating-tension module extends to nematic junctional forces, motivating the transfer to phase-field models."},{"cited_title":"Mueller, J","cited_arxiv_id":null,"evidence_quote":"Provides the multi-phase field equations of motion and free energy that the model modifies."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplemental Material; provides the finite-time checks (MSD exponents, early-tau subrange) that defend the non-monotonic claim against run-length artifacts."}],"review_version":1}